Most restaurants spend to find new guests and almost nothing to keep the ones they already served. Email and SMS are the cheapest way to close that gap — if you collect the list properly and send messages people actually want.
August 4, 2026
Here is a pattern that shows up in almost every restaurant we look at. Marketing spend goes almost entirely toward finding people who have never eaten there — paid social, a delivery marketplace promotion, a local influencer meal. Meanwhile the guests who already came in, already liked the food, and already paid full price walk out the door and are never contacted again. You know more about a person who ordered from you last Tuesday than any ad platform does, and in most kitchens that knowledge dies at the terminal. Email and SMS are the cheapest tools available for fixing that. They are also the easiest to do badly, so this is a practical guide to doing them properly.
The argument is not that email is exciting. It is that the economics are lopsided in a way nothing else matches. A marketplace promotion costs you commission on every order plus the discount. A paid social campaign costs money every single time it runs, and the audience is rented — stop paying and it vanishes. An email list costs a small monthly fee regardless of how many times you use it, and you own it. SMS costs a fraction of a cent per message. If a campaign brings back even a small share of a few thousand past guests, the arithmetic tends to work out heavily in your favour.
The second argument is targeting. You are not guessing at a demographic. You are messaging people who have literally handed you money before, and if your systems are joined up you know what they ordered, how often, how much they spent, and when they last came in. That is the difference between a blast and a reason to visit.
Nothing else in this guide matters if the list is 200 people. List building is an operational habit, not a campaign, and the best sources are the moments where a guest is already interacting with you:
One warning: a list spread across six disconnected tools is barely a list at all. If your reservations, ordering, loyalty, and POS each hold a fragment of the same guest, you cannot segment and you will send someone a we-miss-you message the day after they dined with you. This is the practical case for keeping guest data in the platform that already runs your service rather than exporting spreadsheets between systems.
Operators often treat these as interchangeable channels with different character limits. They are not. SMS is interruptive, nearly always read, and expensive in goodwill; email is patient, roomier, and cheap. Use each for what it is good at.
| SMS | ||
|---|---|---|
| Best for | Storytelling, menu launches, longer offers, newsletters | Time-sensitive nudges: today only, last two hours, order deadline |
| Typical frequency | Two to four times a month | One to three times a month, at most |
| Cost per send | Effectively free at restaurant list sizes | Small per-message fee that adds up at scale |
| Tolerance for a bad send | Moderate — people ignore it | Low — people unsubscribe or get annoyed |
| Consent standard | Explicit opt-in, clear unsubscribe | Stricter: separate, unambiguous opt-in for marketing texts |
A reasonable default is email as the workhorse and SMS reserved for genuine urgency and for your best customers. The fastest way to burn an SMS list is to use it for things that could have waited until Thursday.
This is the part operators skip and later regret. In the United States, marketing text messages fall under telemarketing rules that require prior express written consent, which in practice means a clearly labelled opt-in that the person actively took — not a pre-ticked box, and not a phone number they gave you to receive an order update. Commercial email requires accurate sender information, a working unsubscribe, and honouring opt-outs promptly. In Canada, the anti-spam legislation is stricter still: you generally need express consent, you must identify yourself, and there are limited implied-consent situations with time limits attached. Penalties in both countries are real, and the enforcement risk sits with you, not with your software vendor.
The practical version: use separate checkboxes for order updates and for marketing, record when and how each person opted in, make unsubscribing genuinely easy, and process opt-outs immediately across both channels. If you operate on both sides of the border, run to the stricter standard everywhere — it is simpler than maintaining two behaviours. None of this is legal advice; if you are running large campaigns, have someone qualified look at your opt-in flow once.
You do not need a content calendar. You need a handful of messages that trigger themselves and a small number of campaigns you actually plan. Start here:
Notice how few of these are discounts. Discounting your regulars is the most common mistake in restaurant CRM: you take margin off people who were going to visit anyway. Reserve the money for genuinely lapsed guests and genuinely slow shifts.
Lists die from three things, in roughly this order. Sending too often with nothing to say, so people stop opening and eventually leave. Sending the same generic blast to everyone, so it never feels relevant. And going silent for four months and then reappearing with a promotion, which reads as desperation and drives unsubscribes.
A steady, modest rhythm beats bursts. Pick a frequency you can sustain in your slowest week and hold it. Write like a person — the subject line that works is usually the plain one that says what the message contains, not the clever one. And watch your unsubscribe rate more carefully than your open rate: opens are noisy and increasingly unreliable, while unsubscribes tell you honestly whether people want to hear from you.
Opens and clicks are activity, not results. The number that matters is attributed revenue: of the people who received a given campaign, how many transacted with you in the following week or two, and what did they spend. That is only measurable if your marketing tool and your point of sale are the same system or are properly joined — otherwise you are comparing a click report to a sales report and hoping.
This is one of the underrated arguments for consolidating tools. When ordering, loyalty, and marketing sit on the same platform as the POS, the loop closes automatically: the segment is built from real purchase history, the message goes out, and the resulting orders are attributed back without anyone exporting a CSV. If you are running multiple locations, that also means one guest profile across the group rather than five separate lists that each think they know the same customer.
Two related reads if you are building this out: our guide to setting up direct online ordering, which is where most of your list will come from, and the loyalty program guide, which covers the mechanics of turning contact details into repeat visits.
Month one, fix collection: digital receipts on at the terminal, signup on the ordering site, a QR on takeout bags, and the team briefed on how to ask. Month two, turn on the automated messages — welcome, birthday, lapsed — and leave them alone. Month three, start one planned send a month and one targeted push at your weakest shift, and measure both against revenue rather than opens. That is a complete programme, and it is roughly a day of setup plus twenty minutes a month afterwards.
Smaller than most operators assume. Even a few hundred past guests is enough to make an automated welcome and lapsed-guest message worth switching on, because the cost of sending is effectively zero. The list grows fastest once digital receipts and direct online ordering are collecting contacts on every transaction, so start the collection habit immediately rather than waiting until the list feels big enough.
Not for marketing, unless they separately and clearly opted in to marketing messages. A number given to receive an order status update is consent for that transactional message only. Use a distinct, unticked checkbox for promotional texts, record when and how it was given, and process opt-outs immediately. Rules are stricter in Canada than in the US, so if you operate in both, run to the stricter standard.
No, and over-discounting is the most common way restaurants lose money on this channel. News sends — a new menu, a holiday order deadline, a patio opening — often outperform offers and cost you nothing in margin. Save real discounts for guests who have genuinely lapsed and for shifts that were going to be slow anyway.
Match recipients against sales in the following one to two weeks and compare against a group who did not receive it. That is straightforward when your marketing and point of sale are the same system, and close to guesswork when they are separate tools. If you are evaluating vendors, ask specifically how campaign revenue is attributed back to orders.